SEO KPI Reporting: Which Numbers Survive at Your Deal Volume

The short answer

Report the numbers that survive your deal volume, and report counts before ratios. For a B2B company closing a handful of deals a month, the useful SEO KPI set is small: booked sales conversations from organic, commercial-page sessions, impressions on the queries you actually sell against, and your presence in AI answers split into citations and brand mentions. The popular replacements for rankings and traffic, meaning SEO ROI, cost per acquisition and LTV to CAC, are ratios. Ratios need a denominator, and at low volume yours is too small to mean anything.

That is the part the standard framework skips. Nearly every article on this keyword runs the same shape: an executive asks what SEO generated in revenue, the room goes quiet, then the author prescribes eleven or twenty KPIs sorted by audience. The prescription assumes a CRM, multi-touch attribution and enough closed deals to build cohorts. If you close five deals a month and you are still the closer, that framework hands you a confident dashboard built on four data points.

Rankings, impressions and traffic can all rise while nothing books

The vanity-metric framing is popular and it is half right. Calling rankings and traffic shallow misses the actual failure. Google no longer passes them through in a predictable ratio, so the same movement now has several possible causes and your report cannot tell them apart.

The click side has moved fast. In the first four months of 2026, 68.01% of US Google searches ended without a click, according to SparkToro's analysis of Similarweb clickstream data. That figure was 60.45% in 2024, a 7.5 percentage point move in two years. AI Overviews now appear on more than 20% of all searches, and when one is present SparkToro puts the reduction in click-through at close to 60%.

Then there is the measurement problem underneath it. Google's own documentation on AI features and your website states that sites appearing in AI Overviews and AI Mode are included in overall search traffic in Search Console, reported within the Web search type. They are not broken out. So the one report you would open to diagnose a flat month blends classic result exposure with AI-feature exposure into a single impressions line.

The practical effect is that impressions can climb because Google started showing your page inside an answer box that resolves the question on the page. Rankings can improve on queries that no longer produce visits. Both numbers go up. The calendar stays empty. Nobody in the report is lying, and nobody in the report can explain it either. If AI Mode exposure is a growing share of your footprint, how AI Mode changes what ranking means is worth reading alongside your next report.

The CTR benchmark you are failing is mostly a site-size artifact

Here is a specific way KPI reports mislead small companies. Someone compares your organic click-through rate against a published benchmark, finds you below it, then books a quarter of work on fixing title tags.

Ahrefs pulled anonymized Search Console data from more than 400,000 websites and broke median organic CTR down by site size. Sites with 0-10 indexed pages ran a 0.58% median. Sites with 5,000 or more ran 1.74%. The number rises steadily across every band in between. Part of that is authority, and part of it is that large sites rank for far more branded and navigational queries, where the searcher was already looking for them by name.

So a blended CTR benchmark is largely a measure of how much branded demand a site already has. A 30-page site that sells to B2B buyers is being compared against companies whose traffic includes people typing their name. Reading your gap against that average as a content quality problem is a misread of the denominator.

The same study found AI Overviews on 57.9% of all question queries, with 99.9% of AI Overviews appearing on informational intent. The more your published work skews toward how and what and why, the more of your impressions sit behind an answer. That is a reason to weight commercial-intent coverage in your reporting rather than a reason to rewrite headlines, and it is the same filter that should have run before the content was commissioned. Our post on why a keyword gap report is not a content plan covers that scoring step.

A better version of this KPI is your own trend rather than an industry average. Split your query set into informational and commercial. Track impressions against clicks separately for each, over a rolling 90 days. If informational impressions rise while informational clicks stay flat and commercial clicks hold, that is answer absorption doing what it does, and it is not a failure you can fix by editing. If commercial clicks fall while commercial impressions hold, that is a real problem on pages that matter.

Ratio KPIs need a denominator you probably do not have

This is the gap in every framework on this topic, and it is the one that costs the most.

Take a company booking five sales conversations a month from organic. Compute a monthly conversion rate on that and one extra booking moves it 20 percentage points. Compute LTV to CAC on four closed deals in a quarter and a single deal closing in week thirteen instead of week one changes the answer. The number is arithmetically correct and it carries no information about whether anything improved. Two months of movement in a metric like that is noise wearing a decimal point.

I ran marketing for a company that made the Inc. 5000 four years running, taking it from startup to exit. The most expensive reporting mistakes I saw there were never missing dashboards. They were confident month-over-month percentage moves on denominators of four or five, presented to people who then reallocated real budget against them.

A working rule. If one additional event would move a percentage by more than about two points, report the count instead of the percentage. Then report the percentage on a trailing window long enough that the denominator survives one deal moving, which at these volumes usually means 90 days minimum and often two quarters.

What that looks like in practice for a company at this scale:

Report as counts, monthly: booked sales conversations from organic, held conversations, sessions on commercial-intent pages, impressions on your commercial query set, and pages newly earning any AI-answer presence.

Report as ratios, trailing 90 days or longer: commercial-page session to booked conversation, booked to held, held to opportunity, then cost per booked conversation once there is a spend figure worth dividing.

Do not report at all until the denominator exists: LTV to CAC, blended SEO ROI, cost per acquisition, any cohort-level conversion rate. Standing these up early does not buy rigor. It buys a number that argues with itself every month.

Run the math forward, not backward

Every page on this keyword reports backward. Here is what happened, now tie it to revenue. The calculation a founder actually needs runs the other way, and it takes about ten minutes.

Start from customers and work down to impressions. Say you want four new customers this quarter. You close one in four booked sales conversations, so you need sixteen booked conversations. On your commercial pages, roughly 1.5% of sessions turn into a booking, which puts you at about 1,067 sessions on those pages across the quarter, or roughly 355 a month. At the 0.58% median CTR for a site your size, 355 clicks a month needs somewhere near 61,000 monthly impressions on commercial queries.

Now compare that to your current Search Console number. If you are sitting at 4,000 impressions a month on commercial queries, no KPI framework closes that gap this quarter, and no amount of reporting cadence changes the answer. The honest read is that search is a compounding asset that will not carry the quarter, so something else has to fill the calendar while it builds. That is why we sequence outbound first and let search and AI-search visibility compound underneath it, with organic pipeline expected from around month four rather than month one.

If you are at 45,000 impressions, the gap is conversion path and coverage rather than demand, and the work sits on your money pages. If you are past 61,000 and still short on bookings, the leak is downstream of the click, which is where an audit ordered by what each failure costs you earns its fee.

The value of running it forward is that it converts an argument about metrics into a single question with a number attached: is the current trajectory in the ballpark of the target, yes or no. That question has an answer. "Did SEO drive revenue" does not, at four closed deals.

Two numbers most reports still miss

Being cited by an AI engine and being named by it are different events, and reports that collapse them into one AI visibility figure hide the more valuable half.

Semrush, working with Kevin Indig, logged 3,981 domain appearances across 115 prompts in 14 countries and four AI search engines. 61.7% were ghost citations: the page was used as a source link and the brand name never appeared in the answer. Only 13.2% were both cited and mentioned. The engines also behave differently enough that a blended number is misleading. ChatGPT cited at 87% and mentioned brands 20.7% of the time. Gemini did close to the reverse, naming the brand 83.7% of the time while citing the source only 21.4%. One finding translates straight into a content decision: comparative content produced 2.4 times more brand mentions than informational content.

So track citation rate and mention rate as two lines, per engine. A page can be feeding answers all quarter while every reader walks away without your name, which shows up in your report as AI visibility and in your pipeline as nothing. Tooling for this has matured enough to be worth a look, and we compared the options in the best AI search visibility tools.

One more figure worth putting in front of whoever owns the budget. In Semrush's 2026 AI Visibility Index, built on 126 million US AI search prompts plus a companion survey, 81% of organizations that ran SEO and AI visibility as one workflow reported increased traffic or leads from AI platforms, against 36% of those managing them separately. That is an argument for one reporting surface covering both, rather than an AI visibility slide bolted onto the end of an SEO deck.

What the report should say when a number moves

A KPI report earns its place when every line has an owner and a decision attached. For each number, write down in advance what you would do if it moved 20% in either direction. Anything with no answer to that comes off the report. Most dashboards fail this test on the majority of their tiles, which is why nobody reads them by month three.

Keep one number at the top that the whole system is accountable for. We use booked sales conversations, because it is the last point where marketing and sales share a definition and the first point where the founder's calendar is affected. Everything else on the page exists to explain a move in that number.

Then say the quiet part in the report itself. Which lines are counts and which are ratios. Which ratios do not have a denominator yet. Which impressions figure includes AI-feature exposure you cannot isolate. A report that states its own limits is more useful than one that presents four data points as a trend, and it is the version that survives a founder asking a hard question about it.

If you want an outside read on which of these numbers your Demand Engine can currently produce, and where the path from impression to booked conversation is actually breaking, the Growth System Score is a 48-hour diagnostic video that grades all three engines and ranks every leak by what it is costing you. No call required.